In my AT Sprint, I have at least one of my Q3/Q4/Q5 2N7000's acting up, 
and staying "on" all the time, which in turn burns up Q7, the BCP69CT.  
Odd thing is, the gate waveform is still good. 

I cannot readily put my hands on the PNP power switch Q7 BCP69CT - 
could you send a replacement? 

I have spares for the 2N7000's... but also a question: do I need to match Vgth?

I wonder if you've gotten a handle on the odd "noise" problem you mentioned 
during power-up, and I wonder if that's what might be causing the finals to 
crap out?  You said to disconnect the antenna when powering up, and I confess 
leaving the dummy load on while powering up, so maybe that is what caused my 
problem.  Can you explain what's going on there?  

My theory - weak though it is, because I haven't actually witnessed the 
failure mode - is that uP pin 2 (PD0) is defined as open-collector during 
reset and power-up, leaving U5b free to go or float "low", making the final's 
gate "high".  Perhaps there's a way to "gate-out" the gate signal during the 
power-up sequence.  I know nada about the uP, but I wonder if pin 2 (PD0) had 
a hefty pull-up resistor on it, it might assure the finals' gate stayed low 
during the power-up period when PD0 might be undefined (i.e., O/C)

I'm really enjoying the build so far...

monty  N5ESE